fedi-feed-router/backend/src/Domains/Article/Parsers/Vrt/VrtHomepageParser.php

<?php

namespace Domains\Article\Parsers\Vrt;

class VrtHomepageParser
{
    /**
     * @return array<int, string>
     */
    public static function extractArticleUrls(string $html): array
    {
        // Extract article links using regex
        preg_match_all('/href="(\/vrtnws\/en\/\d{4}\/\d{2}\/\d{2}\/[^"]+)"/', $html, $matches);

        $urls = collect($matches[1])
            ->unique()
            ->map(fn ($path) => 'https://www.vrt.be' . $path)
            ->toArray();

        return $urls;
    }
}